Recovery after treatment and sensitivity to base rate

Accurate classification of patients as having recovered after psychotherapy depends largely on the base rate of such recovery. This article presents m ethods for classifying participants as recovered after therapy. The approac h described here considers base rate in the statistical model. These method s can be applied to psychotherapy outcome data for 2 purposes: (a) to deter mine the robustness of a data set to differing base-rate assumptions and (b ) to formulate an appropriate cutoff that is beyond the range of cases that are not robust to plausible base-rate assumptions. Discussion addresses a fundamental premise underlying the study of recovery after psychotherapy.